Good critical take on the original GeekyAnts piece. I especially liked the point about enterprise AI becoming an ongoing operating capability rather than something you simply “make production-ready” once. The focus on continuous evaluation, cost per successful task, and human oversight makes the discussion much more practical.